The objective of this proposal is to develop a medical information processing system that can perform the following: (1) provide transparent and easy access to the complete electronic patient record, and (2) retrieve medical documents and images based on information content (e.g., pathology, radiology finding) rather than artificial keys such as patient hospital identification number. A flexible six-level information processing architecture is proposed to perform high-level query answering functions. Central to the project is a three-level object-oriented data model to capture the semantic contents of medical documents and images. The data model will incorporate the National Library of Medicine's Unified Medical Language System (UMLS) Medical Semantic Network and Metathesaurus. Powerful computer-aided software engineering (CASE) tools are used to perform graphical data modeling and rapid code development for the integration of legacy medical information systems. Testbed systems for thoracic and neuroradiology will be developed to demonstrate the clinical utility of the system. A highly graphical multimedia query language is introduced as the system's user interface.